Ontology-driven representation of knowledge for geological maps

https://doi.org/10.1016/j.cageo.2020.104446Get rights and content

Highlights

  • Ontological approach is needed to reduce the ambiguity of geological concepts.

  • Concepts and terms deriving from geosciences standard vocabularies have been encoded in the Ontology (OntoGeonous).

  • Ontology-Driven geodatabase implementation reduces the ambiguity of geological maps.

Abstract

This paper presents an ontology-driven representation of knowledge for geological maps. The ontological formal language allows for a machine-readable encoding of the Earth scientist's interpretation through semantic categories and properties and is credited to support knowledge sharing and interoperability. We introduce an ontology-driven method for the interpretation and the encoding of the map data that employs shared vocabularies and resources encoded through ontologies in order to prevent the use of ambiguous terms. The approach relies on a computational ontology of the geological knowledge (OntoGeonous), which formalizes a number of geological knowledge sources (including GeoScienceML), to guide the interpretation process. The design of the database underlying the map (OntoGeoBase) constrains the process of data entry to refer to the terminology conveyed by the taxonomic-axiomatic nature of the ontology. This reduces the amount of implicit knowledge favouring a conceptual alignment of the ancillary documentation with the map, leading to a better comprehension of map and allowing the traceability of the interpretation.

Introduction

In the last two decades, the major methodological innovation in the production of geological maps has essentially concerned the usage of tablets and small size PC's, equipped with a GIS software, directly on the field (see, e.g., McCaffrey et al., 2005; Pavlis et al., 2010; Whitmeyer et al., 2010, DeDonatis et al., 2016). This technological advance favoured the acquisition of information in a digital format since the beginning of the mapping process in the field, in a suitable way for data reworking and sharing. However, the remarkable advantages brought about by these IT innovations have mainly addressed the digital recording and representation of spatial data. Some important features, such as the interoperability and the unambiguity of data, has not been addressed by these new technologies. This paper focuses mostly on the final step of the digital mapping process, i.e. the representation of geological knowledge in maps and in particular the disambiguation of concepts through the semantic formalization. Semantic formalization of concepts is conceived to address the peculiarities of the domain, allows abstracting objects from the real world to the information world, while the implementation of the resulting database will be driven by the logical constraints provided by a computational ontology.

The semantics-informed design of the database through a computational ontology, although largely debated in the literature (Uschold, 2015), fits the problems of geological knowledge representation in maps, since it: (i) focuses on the formal definition of classes (or categories, where the instances are members of classes; (ii) provides strong constraints (axioms) to convey data meaning, together with categories, for consistency and reasoning; (iii) comes with reasoning algorithms, to infer new information. In fact, the ontological approach is valuable when the encoding of human thinking is a crucial issue with respect to data encoding (Uschold, 2015). We claim that the importance of the terminological issue as well as the engagement of the Earth scientists in the design process is paramount in geological mapping, especially in the light of terminological effort promoted by the international committees that goes together with the ontological development (more on this below).

It is known that geological mapping largely consists in a process of inferencing (Brodaric, 2004; Balestro and Piana, 2007; Loudon, 2009, 2011) because, in geology, rocks often document events that are inferred to have occurred, and not directly observed1. Consequently, from the observation of rocks and landforms on the field to the production of a geological map, many decisions are taken, and many of them are influenced by pre existing models of the geological evolution of the map area. Moreover, in the geological maps much of the knowledge is implicit (tacit knowledge, such as fundamental principles, intended meanings and assumptions). This working method may cause a loss of reproducibility of the data, because of the difficulty to separate data from interpretations, in contrast with the paradigm of the self-correcting nature of science. Therefore, we believe that the ontological approach, with its formal and explicit representation format, can effectively guide the representation of geological knowledge on maps. An ontology-consistent description of the mapped features requires to make explicit much information (namely, classes, properties, and axioms) and leads to a retractable path of interpretation. The ontology axioms can unambiguously encode the relationships of the geologic features with some Geologic Event, which is the key to the reconstruction of the geological history of the map area. All such explicit information is expressed in a machine-readable language, which allows for the automatic inference (reasoning) and the consistency checking.

In this paper, we employ ontologies for the formal representation of geologic knowledge and the consequent conceptual design of the database schema to address explicitly the interpretation of the mapped features. As a proof of concept, we also describe how to translate the conceptual schema into a logical database schema through a well-known GIS software. Further, the use of computational ontologies improves the employment of shared vocabularies, which in turn support interoperability and data sharing. Nowadays, effective data sharing through the reference to a common framework (e.g., GeoScienceML2) is still rarely supported by geology data infrastructures. Several initiatives aimed at providing a knowledge infrastructure for geosciences addressed a number of issues in the literature. Certain approaches have provided wide scope analyses, which have sketched the scenarios of the infrastructure, such as business models, the assessment of the needs, the formal specification of the requirements (Buller, 2005; Brodaric and Gahegan, 2006; Raskin, 2006) Reitsma et al., 2009; Loudon, 2011). Other approaches have provided some concrete implementations, from the early database schemata (Laxton and Becken, 1996) to the tools for the collection of field data (Dey and Ghosh, 2008) and the definition of standard vocabularies for the harmonization of terminology (Raymond et al., 2016). Many of these initiatives have promoted the usage of ontologies as the major tool for the maintenance of the knowledge assets within the geoscience community or the survey organizations (Howard et al., 2009) or for addressing the issue of data heterogeneity (Abel et al., 2015) and have developed ontologies for limited domains, such as, e.g., field activities (Brodaric, 2004; Hwang, 2012; Boyd, 2016), geochronological periods (Ma et al., 2011), lithological materials (Richard, 2006; Sinha et al., 2006; CGI SimpleLithology3), algorithmic interpretation of sedimentary facies for the individuation of geologic processes (Carbonera et al., 2015).

Recently, we have been developing a logical framework for joining the efforts of the rigorousness of the ontological approach with the standardization of vocabularies, and we have applied the knowledge base to inform the terminology of a geological mapping process (Piana et al., 2017a, Piana et al., 2017a; Lombardo et al., 2018). The method used here for the description and sharing of the geological knowledge in a map is to leverage on ontology OntoGeonous4,5 (Lombardo et al., 2018), which axiomatizes the vocabularies, UML schemata and natural language definitions provided by GeoScienceML and other knowledge sources. We have designed a geodatabase, named OntoGeoBase, in which the process of data entry is terminologically constrained through the ontological terms.

In line with other approaches that pursue the alignment of representation (such as, e.g., Cox and Richard, 2015), we address the general representation of geological knowledge by encoding the general statements reported in the international standard documentation (such GeoScienceML, INSPIRE, SWEET) and by encoding the specific statements related to the geological map in a consistent way with respect to such general statements. Since the realisation of a geological map is a synthesys process that usually requires many decisions and choices among different interpretative solutions, we believe that an ontological approach grounded on a robust semantic knowledge could allow for a reduction of ambiguities and/or implicit knowledge. We claim that the adherence to the international standards puts our approach in a wider perspective in terms of reusability and interoperability.

The paper is structured as follows: in Section 2 we report about the current state of the art, i.e. how digital geological mapping has become a common practice in the last decades, and how it shaped, in some countries, the operative framework of the national mapping projects. Section 3 presents the background technologies for our approach, namely the existing standard vocabularies and their encoding into the ontology OntoGeonous. Section 4, the core of the paper, reports on how the ontology can account for the interpretation of the geological knowledge to be reported in the map. As a proof of concept, Section 5 describes how we can shape the OntoGeoBase schema from the ontology axioms and properties. Section 6 discusses the novelties and the impact of our approach on the geological mapping process. Finally, Section 7 reports out conclusions.

Section snippets

State of art in geological mapping and vocabularies

In this section, we address the state of the art in the current methods of digital geological mapping, its standardization, and the role of shared vocabularies. With the expression “Digital geological mapping”, we intend all the processes that lead to creation of a map, from the beginning of the working process (and directly in the field) to the sharing of the geological map and its contained knowledge. In this paper, however, we focus on the representation of the data contained in a geological

OntoGeonous and the ontologies for the geosciences

Computational ontologies has proved to be effective in sense disambiguation (see, e.g., Navigli and Velardi, 2005). Ontology OntoGeonous concerns the definition of the geological features, which are encoded into formally axiomatized classes, including properties between concepts (see also Lombardo et al., 2018). Here we review ontology OntoGeonous and the major properties employed in the representation of geological knowledge in maps.

Ontologies usually provide the semantic backbone for the

OntoGeonous for the geological map representation

The geological map is a synthesis and interpretation of a wide range of data sources (Harrison, 1963), implemented in the frame of interpretative and historically based geological reasoning (Frodeman, 1995). Different vocabularies and concepts are to be considered, depending on which kind of geological features we want to highlight in the map. One of the relevant difficulties in understanding geological maps relies on the fact that they are actually four-dimensional data systems, where the

Proof of concept: from OntoGeonous to OntoGeoBase

In this section, we apply the ontology classes, axioms, and properties to the design of a database logical schema, named OntoGeoBase, and we present its implementation into a well-known GIS application.

OntoGeoBase is a relational database, i.e. a number of tables (the so-called relations) consisting of rows and columns. Each row represents an entity (or instance, in ontological terms) and the columns represent the attributes of such entity; the whole table represents an entity type (or class).

Summary of the method and discussion

Finally, we sum up the method we have implemented to represent the knowledge underlying a geological map through a controlled vocabulary licensed by ontology, claimed as a basic condition for the perspective of data sharing, and discuss pros and cons.

The aim of the paper is to present how the geological knowledge can be represented without ambiguities and in compliance with standard vocabularies through the use of an ontology. The properties in the axioms represent explicit relations over the

Conclusion

In this paper, we propose a method for the representation of the geological knowledge underlying a map, supported by the formal structure of an ontology. Ontologies address human thinking in a formal language and a machine-readable representation, open to reasoning procedures and traceability of information.

The feasibility of the method is demonstrated through the implementation of an ontology-driven design database; axioms, properties, and relations of the ontology become columns in the

CRediT authorship contribution statement

Alizia Mantovani: Software, Data curation, Writing - review & editing. Fabrizio Piana: Conceptualization, Writing - review & editing, Funding acquisition. Vincenzo Lombardo: Conceptualization, Writing - review & editing, Supervision.

Acknowledgements

This research was funded by funds of the IGG (Institute of Geoscience and Earth Resources), Torino Department, of the National Research Council of Italy and by GeoDive Project, University of Torino, Marco Giardino Coordinator.

References (55)

  • F. Reitsma et al.

    Semantics, ontologies and eScience for the geosciences

    Comput. Geosci.

    (2009)
  • M. Sen et al.

    GeoSciML: development of a generic geoscience markup language

    Comput. Geosci.

    (2005)
  • L. Thum et al.

    2D and 3D GIS-based geological and geomechanical survey during tunnel excavation

    Eng. Geol.

    (2015)
  • J. Zhong et al.

    Ontology of fractures

    J. Struct. Geol.

    (2009)
  • M. Abel et al.

    Ontological analysis for information integration in geomodeling

    Earth Sci. India

    (2015)
  • C. Ambrosi et al.

    Improvements in 3-D digital mapping for geomorphological and quaternary geological cartography

    Geograph. Helv.

    (2015)
  • H.A. Babaie et al.

    Designing a modular architecture for the structural geology ontology

    Geoinformatics Data to Knowl. Geological Society of America Special Papers

    (2006)
  • G. Balestro et al.

    The representation of geological knowledge and uncertainty in databases of GIS geological maps

    Boll. Soc. Geol. It. (Ital. J. Geosci.)

    (2007)
  • T. Boyd

    Ontology of Geological Mapping

    (2016)
  • B. Brodaric et al.

    Representing geoscientific knowledge in cyberinfra structure: some challenges, approaches, and implementations

    Geol. Soc. Am. Spec. Pap.

    (2006)
  • B. Brodaric

    Characterizing and representing inference histories in geologic mapping

    Int. J. Geogr. Inf. Sci.

    (2012)
  • G. Buller

    A Conceptual Approach to the Development of Digital Geological Field Data Collection

    (2005)
  • S.J.D. Cox et al.

    A geologic timescale ontology and service

    Earth Sci. India

    (2015)
  • M. De Donatis et al.

    Open source GIS for geological field mapping: research and teaching experience

    Peer J Prepr

    (2016)
  • R. Frodeman

    Geological reasoning: geology as an interpretive and historical science

    Geol. Soc. Am. Bull.

    (1995)
  • J.M. Harrison

    Nature and significance of geological maps

  • E. Hintersberger et al.

    The new database “tectonic boundaries” at the geological survey of Austria

    Jb. Geol. B.-A.

    (2017)
  • Cited by (0)

    View full text